Nirvana Concepts - Performance, Scalabilty and Resilience
Performance, Scalability and Resilience are design themes that are followed in all areas of Nirvana's design and architecture. Specific implementation features have been introduced into server and client components to ensure that these themes remain constantly adhered to.
Nirvana is capable of meeting the toughest of Low Latency and High Throughput demands. Nirvana's server design, sophisticated threading models and heavily optimized IO subsystem ensures peak performance. Multicast and shared memory communication modes allow Nirvana to consistently achieve low microsecond latencies. Nirvana is constantly being benchmarked by clients and other 3rd parties and continues to come out on top.
The benchmarking section provides detailed information on performance in a variety of scenarios. Information on performance tuning is also available on the website, helping clients achieve optimal performance in their deployments.
Scalability in terms of messaging Middleware typically means supporting large numbers of concurrent connections, something Nirvana does out of the box. However in defining truly global enterprise applications a single system often needs to scale across more than one processing core, often in more than one geographic location.
Nirvana servers can be configured in a variety of ways to suit scalability (and resilience) requirements. Multiple Nirvana servers can exist in a single federated name space. This means that although specific resources can be put onto specific Nirvana realm servers any number of resources can be managed and access centrally from a single entry point into Nirvana's federated namespace. In addition to high availability and resilience features Nirvana Clusters also offer a convenient way to replicate data and resources among a number of realm servers thus supporting higher rates of concurrent connections.
Business contingency and disaster recovery planning demand maximum availability from messaging Middleware systems. Nirvana provides a number of server and client features to ensure data always remains readily accessible and that outages are transparent to clients.
Nirvana Clusters replicate all resources between realms. Channels, Topics, Queues and the data held within each one is always accessible from any realm server in the cluster.
Nirvana clients are given a list of Nirvana realms in the cluster and automatically move from one to another if a problem or outage occurs.